How SkySail Algorithms Earn Capital

The lifecycle at a glance

Research: modeling the market as a probabilistic system

SkySail approaches markets the way systems engineers approach any complex, uncertain system: as a set of evolving probabilistic states rather than a price line to be predicted. The research question is not which trade to make. It is what about the system can be measured, and what can be forecast with demonstrable skill.

That question produces foundational models. A foundational model does not trade. It isolates something real in a noisy, stochastic market and makes it usable by the rest of the system. Foundations take several forms:

These roles are not a sequence. Foundations sit side by side, feed one another and combine in different ways, forming an architecture in which every component is proven before anything relies on it. Because they describe how markets behave rather than how one system behaves, most carry across asset classes. Rubix VM, SkySail's foundational risk model, is one of them, and it now operates inside NQ-GX7.

Quantitative execution systems (QES) appear only at the top of that architecture. Each QES is a composite of strategies and algorithms, composed under the GRIP Alpha Model from components that have already proven their skill. Every component must have a mechanism behind it.

Structure and data science: confidence before capital

A backtest answers one question: how well does a model fit the past? Across much of the industry, that answer is treated as proof. At SkySail, it is the beginning of a far harder inquiry, because fit to history and durability in live markets are different properties.

SkySail's structure and data science stage is the proprietary step within M7 that establishes the highest possible confidence a model will hold up live. It does not rely on any single test. It triangulates. A proprietary set of mathematical and statistical characteristics, developed and refined over nearly a decade, is evaluated together, and a model advances only when they agree. The characteristics themselves are proprietary. What they establish is not:

Together, they tell SkySail what will and will not work in live, unseen markets. Development bias can never be eliminated entirely, but M7 reduces it as far as discipline and science allow.

Live deployment: the record begins on day one

Deployment is the point at which a model begins trading real capital in real brokerage accounts. At SkySail, "launched" means exactly that. A model's track record begins on its launch date, and every figure in it reflects live execution, net of commissions and transaction costs.

NQ-GX7, SkySail's current QES, trades Nasdaq-100 index futures and was deployed in June 2024. Its live record is independently audited against broker statements from that month forward.

A credible live record shows its stress as clearly as its returns. SkySail publishes NQ-GX7's complete drawdown history alongside its performance.

Capital allocation: every model competes for its place

Deployment does not end the evaluation. Every model in the portfolio competes for the same three finite resources: capital, research time and risk budget. Allocation flows toward the models delivering the strongest risk-adjusted results, and away from those that are not.

Risk management accounts for roughly 90 percent of SkySail's research and process effort. Maximum exposure for every position is governed relative to the account, which makes position size an output of risk rather than of the leverage futures make available.

Live models are monitored continuously against the characteristics they were engineered to deliver. A model that drifts from its intended behavior loses allocation.

Retirement: profitable is not sufficient

Understanding where a system stops working is as important as understanding where it works. SkySail studies the limits of its models as rigorously as their strengths, and that understanding governs every retirement decision.

A model is retired when it no longer keeps pace with the rest of the portfolio, not merely when it stops making money. Capital, research time and risk budget belong with the models that put them to the best use.

The Sky III example

Sky III (also written Sky3 or Sky 3) was a forex QES deployed in 2022. It was delivered through SkySail's non-custodial model, in which clients kept their capital in their own brokerage accounts, and several SkySail entities, including SkySail Strategies and SkySail Investments, had access to it.

By 2024, SkySail's newer models, including NQ-GXT and NQ-GX7, were producing stronger risk-adjusted returns. Sky III remained profitable, but it no longer kept pace with the rest of the portfolio. Its allocation was reduced over time, and in September 2024 SkySail retired it and stopped offering it.

One system, one record

When a model is retired, what it taught SkySail remains with the firm. Its record remains with the model.

Sky III's results are not combined with, extended into, or presented as part of NQ-GX7's track record or any other system's. Reviews and posts from 2022 to 2024 that mention Sky III refer to that retired system, not to anything SkySail offers today.

SkySail began in 2019 as Agile Capital Solutions, which merged into SkySail Strategies in 2023, and operated two UK companies, including SkySail Investments Ltd, from 2024 until it voluntarily dissolved them in 2025.

Frequently asked questions